In a principal - multi-agent relationship, we derive the optimal mutualmonitoring - incentive pay mix. When the agents are better informed about theireffort choices than the principal, and when their information is sufficiently"good" there is a substituability between those two modes of providingincentives. However we show that the optimal mix depends on agents' liabilitylimit. When it is sufficiently slack the principal uses stronger incentive pay andless mutual monitoring. We also derive the conditions for adoption of costlymutual monitoring technology.